Speaker
Description
Language Models’ performance has proven to be directly related with the number of parameters the models have. There is an international race to compute performant models for all the languages and domains as this will invigorate all the processes that rely on Natural Language. It has been recently shown that domain-specific models are as performant as large scale models while domain-specific models are smaller and easier to manage.
As computational capacity and data amounts are allowing domain-specific Language Models to emerge, it is demanding for domain professionals to understand the capabilities and limitations of Language Models as well as their evaluation procedures among other questions.
In this presentation, we provide a brief explanation on Language Models and data gathering methods. Then, we will cover domain-specific Language Models and their relevance and current use cases in Humanities. Finally, we will cover ethical, bias and surveillance concerns derived from the usage and application of these Language Models.
